#cheriot

I have had a tab open with the JavaCard spec for a few weeks and I continue to resist the temptation to write a JavaCard implementation for CHERIoT. Please be impressed. I honestly didn't think I'd manage this long. #CHERI #CHERIoT #JavaCard #Java

My latest article for ACM Queue about CHERIoT is live! This describes the design decisions that we made in designing an aggressively privilege-separated OS with a single address space for embedded systems. It's part of a series of articles about designing single-address-space systems and aims to highlight which parts of CHERIoT are things that are a good idea everywhere and which are a good idea only for embedded systems. EDIT: Looks as if they mangled the title, it should read 'usable' not 'able'. EDIT2: And now they have fixed it. #CHERIoT #CHERI #ACMQueue

We're going to be hiring someone to work on CHERIoT software things soon. Full remote, but UK Only (for various funding purposes, we need more people in the UK). Happy to talk to junior people who have some C/C++/Rust experience. You'll initially be mostly writing C++, though we expect this to move to more Rust over the next year or so. Some of the job will be driver development gruntwork (here is a datasheet, here is a software interface for this device class, please write a driver for the device that exposes the interface), some will be at a higher level of the stack (so, how do you design a secure, compartmentalised {something} stack?). If these sound interesting, let me know! #GetFediHired #CHERIoT

I guess, in light of the liblzma debacle, today is a good time to remind everyone that the #CHERIoT platform was designed from the ground up with supply-chain security in mind. If you want to use some third-party code, you can audit precisely the APIs from other components that it can use, the set of things that can call it, the set of devices it can directly access, the amount of heap memory it can allocate, and more. For a case study, see our ongoing work on compartmentalising the network stack, where we can fearlessly reuse third-party code and know that we are safe from entire classes of compromise. If you want to build IoT devices with long, low-maintenance, secure lifetimes, SCI Semiconductor may have the microcontroller that you need to realise your goals.

@fanf@mendeddrum.org Follow-on work is ongoing, focused on #CHERIoT.

I wrote this post over a year ago, but it probably needs repeating periodically. From my perspective, CHERI isn't really about security, it's about making it possible to solve a bunch of interesting language-interoperability and end-user-programming problems that happen to need a bunch of security fixed as building blocks. And so we get a load of security stuff for free. But that's not why I worked on it. #CHERI #CHERIoT

When we originally created #CHERIoT, we added an instruction to make accessing globals uniform with respect to the rest of RISC-V. We fairly quickly realised it wasn’t a great design, but the toolchain changes required to eliminate it took a while. We’ve finally done it and I spent a bit of time this week writing up our journey. Removing this instruction is something we always planned to do before rebasing on the upcoming RV32YE base. #CHERI #RISCV